Download your lesson workbook · Open the assignment tool →
Your ASO learning objective
Collect candidate terms with enough provenance to reconstruct where each idea came from and why it deserves investigation.
- Step 1Preserve the original source
- Step 2Map columns and inspect sample rows
- Step 3Import pending candidates with provenance
Build a source ledger, not a bag of words
Start with product language, customer interviews, support questions, reviews and observed store listings. Each source answers a different question. A support ticket can reveal confusing terminology; a competitor headline can reveal positioning; an imported keyword estimate can suggest demand within the provider’s method. None automatically proves that a term belongs in your metadata.
For every candidate, preserve the original wording, source reference, locale and collection date. If you shorten a phrase later, the original should remain visible. “Remember my walks” becoming “walking diary” is an editorial interpretation, so explain the transformation rather than making it look like a direct customer quotation.
Import deliberately and inspect the mapping
Use a CSV with a header row when you have structured research. Map the keyword text column first, then source, locale, intent, cluster and any metric fields. Check the first three mapped rows before importing. A column named “difficulty” should not become a relevance judgment; put its supplied value in the metric field and explain its provider and scale.
The Workbench appends records and preserves the raw source import in the backup. It flags normalized duplicates without deleting them because two sources may support the same phrase independently. For a focused first exercise, start with 15–30 candidates. This is a practical review batch, not a claim that an optimal keyword set has a universal size.
Make missing provenance actionable
A candidate with no source remains an idea, not a validated opportunity. Record “team brainstorm” when that is the real source. If a date or locale is unknown, keep the gap visible and avoid inventing precision. Before approving a candidate, retrieve the supporting source or document why a product-led judgment is sufficient for the decision at hand.
Worked example: Trail Notes
This fictional hiking-journal app and its review samples are the teaching project. Treat the example as a reasoning exercise you can adapt to your own evidence.
- Original draftA spreadsheet contains “hiking journal”, “hiking journal” and “offline maps”, with no source columns.
- Revised proposalKeep both journal records with their separate interview and listing references. Keep offline maps as a candidate to assess against product scope.
- Why it changesRepeated wording can represent distinct evidence. Importing a phrase does not approve it, and normalization should not erase its provenance.
Your ASO assignment
- Prepare a CSV with term, source, locale, intent and researchedAt columns. Include one quoted cell containing a comma to check the mapping preview.
- Import your candidates. Open duplicate entries, compare their sources and record whether they represent repeated evidence or accidental duplication.
- Export a project backup and confirm that the original source text and original candidate wording remain in it.
Your handover and assessment criteria
Your work is ready for review when each deliverable below can be understood without a verbal explanation. Include the source or product reference, the decision owner and any unresolved evidence gap. If a reviewer cannot verify a claim, record the missing evidence before treating the proposal as approved.
- Deliverable 1Original candidate list and raw import
- Deliverable 2Source, locale and research date per candidate
- Deliverable 3Duplicate review notes and provenance gaps
Check your understanding
Two identical terms come from different customer interviews. Should you automatically remove one?
Read the interpretation
No. Preserve the independent source references. You may consolidate the eventual metadata wording, but do not confuse a unique term count with a count of independent observations.
Record your progress
Mark this lesson complete after checking your assignment against the handover criteria. Progress stays in this browser, separately from your tool project.
Sources and further reading
Platform references checked 30 September 2026.
App Store keyword research guide · Check final metadata fields